Attachment 99488 Details for Bug 79005 – Only discard backing storage on last ref

[patch] Only discard backing storage on last ref

0001-drm-i915-Only-discard-backing-storage-on-releasing-t.patch (text/plain), 2.18 KB, created by Chris Wilson on 2014-05-21 09:26:06 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Chris Wilson

Created: 2014-05-21 09:26:06 UTC

Size: 2.18 KB

patch

obsolete

>From 6b1dd6ee673537803f61b2c5421816418ef418fb Mon Sep 17 00:00:00 2001
>From: Chris Wilson <chris@chris-wilson.co.uk>
>Date: Wed, 21 May 2014 10:21:56 +0100
>Subject: [PATCH] drm/i915: Only discard backing storage on releasing the last
> ref
>
>Before purging our pages (as opposed to copying back the contents from
>the GPU), make sure that there is not an exposed CPU mmapping through
>which the user can inspect the results.
>
>Regression from
>
>commit 5537252b6b6d71fb1a8ed7395a8e5babf91953fd
>Author: Chris Wilson <chris@chris-wilson.co.uk>
>Date:   Tue Mar 25 13:23:06 2014 +0000
>
>    drm/i915: Invalidate our pages under memory pressure
>
>Testcase: igt/gem_mmap/new-object
>Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79005
>Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
>---
> drivers/gpu/drm/i915/i915_gem.c | 20 +++++++++++++++++++-
> 1 file changed, 19 insertions(+), 1 deletion(-)
>
>diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
>index 966a50a1ad8e..9814967f2ed9 100644
>--- a/drivers/gpu/drm/i915/i915_gem.c
>+++ b/drivers/gpu/drm/i915/i915_gem.c
>@@ -4648,6 +4648,24 @@ struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
> 	return obj;
> }
> 
>+static bool discard_backing_storage(struct drm_i915_gem_object *obj)
>+{
>+	/* If we are the last user of the backing storage (be it shmemfs
>+	 * pages or stolen etc), we know that the pages are going to be
>+	 * immediately released. In this case, we can then skip copying
>+	 * back the contents from the GPU.
>+	 */
>+
>+	if (obj->madv == __I915_MADV_PURGED)
>+		return false;
>+
>+	if (obj->base.filp == NULL)
>+		return true;
>+
>+	/* This is racy, but so is userspace racing mmap against close. */
>+	return atomic_long_read(&obj->base.filp->f_count) == 1;
>+}
>+
> void i915_gem_free_object(struct drm_gem_object *gem_obj)
> {
> 	struct drm_i915_gem_object *obj = to_intel_bo(gem_obj);
>@@ -4683,7 +4701,7 @@ void i915_gem_free_object(struct drm_gem_object *gem_obj)
> 
> 	if (WARN_ON(obj->pages_pin_count))
> 		obj->pages_pin_count = 0;
>-	if (obj->madv != __I915_MADV_PURGED)
>+	if (discard_backing_storage(obj))
> 		obj->madv = I915_MADV_DONTNEED;
> 	i915_gem_object_put_pages(obj);
> 	i915_gem_object_free_mmap_offset(obj);
>-- 
>2.0.0.rc2
>

Actions: View | Splinter Review

Attachments on bug 79005: 99476 | 99488 | 99566

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.